Efficient and sensitive detection technologies are crucial for the precise monitoring of food quality and safety. Nanozymes, owing to their superior catalytic activity, can significantly enhance detection signals, thereby effectively improving the sensitivity and accuracy of analytical methods. However, the complexity of food matrices and the multi-dimensional nature of spectral data may compromise the reliability of detection results. Machine learning algorithms possess robust data processing and analysis capabilities, enabling in-depth mining and interpretation of complex detection data, thus effectively improving the accuracy, sensitivity, and efficiency of analytical techniques. This article presents a comprehensive review of the applications of nanozyme-based sensing technologies integrated with machine learning in the field of food quality and safety detection. It particularly highlights the technical advantages of this integration in the detection of food hazards, quality, and authenticity. The combination of machine learning with nanozyme-based sensing technologies not only enhances the detection precision and efficiency but also provides solid technical support for advancing food safety detection toward intelligent and high-throughput systems.

Glucose plays a crucial role in maintaining human health as an indispensable source of energy in living organisms. Accurate monitoring of glucose levels in living organisms and detecting it in food is essential. In this study, gold nanoclusters (AuNCs) with unique aggregation-induced emission (AIE) effects were encapsulated within zeolite imidazole framework (ZIF-8) to fabricate a pH-responsive AuNCs@ZIF-8 fluorescent nanocomposite. The fluorescence intensity had significant sevenfold enhancement compared to AuNCs alone due to the structural domain-limiting effects exerted by ZIF-8, which effectively inhibited the rotations and vibrations of the AuNCs ligands. Based on the increased acidity generated by glucose catalytic oxidation via glucose oxidase (GOx), the subsequent degradation of ZIF-8 structure and the consequent reduction of AuNCs with AIE effects were achieved, and a rapid and efficient fluorescence quantification for glucose was performed. The constructed AuNCs@ZIF-8-based fluorescent probe demonstrated a favorable linear response to glucose, achieving a detection limit of 0.096 mmol/L and providing a rapid and efficient approach suitable for assessing glucose levels in both blood and beverage samples.

Herein, a novel label-free electrochemical immunosensor was fabricated via immobilizing specific anti-β-lactoglobulin (β-LG) antibodies (Abs) onto an integrated electrode of gold nanoparticles (AuNPs)/Prussian blue (PB)/cubic Ia3d structured mesoporous carbon (CMK-8). This immunosensor allowed for the quantitative detection of the major milk allergen β-LG. CMK-8 with excellent electrical conductivity and uniformly adjustable pore structure was modified on the glassy carbon electrode (GCE) and served as the sensitive substrate for the electro-polymerization of PB, forming the redox-active layer. AuNPs were subsequently electrochemically deposited on PB/CMK-8/GCE to improve the electrical conductivity and utilized as the connector for Abs immobilization. During β-LG detection, the Abs-modified AuNPs/PB/CMK-8/GCE exhibited a significant reduction in differential pulse voltammetry current signal when exposed to β-LG, displaying an inverse dose-dependent relationship. The developed electrochemical immunosensor demonstrated good detection performance for β-LG, with a wider linear range of 0.01–100 ng/mL and a lower detection limit of 4.72 pg/mL. Meanwhile, the sensor exhibited remarkable repeatability, reproducibility, stability and anti-interference capabilities, which was further applied to detect β-LG in dairy food, achieving satisfactory recoveries (89.2%–98.8%) and lower relative standard deviation (≤ 3.1%). Therefore, this innovative electrochemical method for food allergen detection holds great potential application in food safety determination and evaluation.

In this study, a label-free, portable and reproducible immunochip based on quartz crystal microbalance (QCM) was developed for the qualitative detection of β-lactoglobulin (β-LG), an allergen, in milk products. Experimental parameters in the fabrication and regeneration procedure such as pH of the coupling microenvironment, amount of anti-β-LG antibody and regeneration reagent were optimized in detail. Under optimal conditions, the proposed QCM immunochip exhibited good recognition of β-LG, with a calibration curve of ΔF = 12.877C_β-LG^0.4809 (R² = 0.9982) and limit of detection of 0.04 μg/mL. Additionally, this portable QCM immunochip had good stability, high specificity, and no obvious cross-reaction to three other milk proteins (α-casein, α-lactalbumin, and lactoferrin). It could compete a qualitative measurement within 5 min, and could be reused at least ten times. In the β-LG analysis of actual milk samples, the developed QCM immunochip yielded reliable and accurate results, which correlated strongly with those from the standard HPLC method (R² = 0.9969). Thus, the portable, stable, and reproducible QCM immunochip developed in this study allowed the rapid, cost-effectively and sensitively measure the β-LG in milk products.